A service level management system where the system learns the structure of a web application, monitors the operation of the application, and controls the processing of incoming requests to achieve optimal performance as defined in a service level agreement. A second reflector is positioned to extend across and face the first reflector with two fluorescent lamps mounted between the two reflectors. An actuator is connected to the feed tube assembly and a receiving tube for advancing and retracting the feed tube assembly to and from the surface of the silicon melt. A control part 3c extending downwardly is disposed at the lower end of the main body part 3. Furthermore, the method includes activating an indicator configured to signal that the predicted amount of time remaining is within the predetermined range of time.